Platform engineering SDLC

AI doesn't fix broken systems - it amplifies them

Lessons from our talk with the DORA Community.

AI doesn’t magically fix broken systems — it makes the cracks impossible to ignore.
That theme surfaced powerfully in a recent talk where the global DORA community, AI-assisted software delivery, and the rise of hyperautomation all came together.

Instead of focusing on tools alone, the discussion dove into how value stream mapping, platform engineering, and organisational culture are reshaping the way modern teams build and ship software. Insights from Davy Kenis, Dick Bevernage, and Nathen Harvey — Lead of the DORA Research Program at Google Cloud — highlighted a shared message: success in the AI era isn’t about doing more, but about doing things differently.

Interested to learn more?

Listen to the full podcast episode or read the article below.

The DORA Community: research that shapes modern IT organisations

The DevOps Research and Assessment (DORA) program is the industry’s benchmark for understanding high-performing software teams. DORA operates as an independent research initiative, analyzing data from thousands of organisations worldwide to identify what truly drives performance, culture, and reliability. Under Nathen Harvey’s leadership, DORA’s work has expanded beyond DevOps metrics to include organisational culture, AI adoption, and developer experience.

“Don’t wait for the report—use the questions as inspiration. Taking the survey already gives teams insights about what to improve,” - Nathen Harvey

Participation in the DORA community is open and free at dora.community, where practitioners can exchange insights and access the latest research findings.

Hyperautomation: a call to work differently

Software delivery has evolved dramatically over the past 15 years—where automation played a crucial role in the past 10 years. However today’s understanding of AI automation is something completely different as we saw before.

We describe today’s acceleration as “hyperautomation”—a shift so profound that it forces organisations to rethink the very foundations of their processes.

Automation isn’t new, but the automation we’re seeing now, at the beginning of the AI era, is completely different. If we don’t start doing things differently, we risk shifting from delivering value to delivering zero value.

Davy Kenis

Principal consultant at AE

Hyperautomation is not about doing the same things faster—it’s about re-designing how organisations collaborate, learn, and deliver business value in an AI-augmented world.

AI as an amplifier, not a shortcut

The 2025 DORA report reinforces that message. It describes AI not as a silver bullet, but as an amplifier: it magnifies both strengths and weaknesses within an organisation.

“AI really has an amplification effect within organisations, if your systems flow smoothly, AI speeds you up. But if you have friction, AI will amplify that pain.” - Nathen Harvey

One clear example is code generation. AI tools help developers produce more code in less time, but if the surrounding software delivery lifecycle (SDLC) is inefficient, the result is simply more work piling up in an already inefficient system.

“If we’re now writing twice as much code as before and we haven’t improved our code-review processes, what we’re really doing is sending twice as much code into what was already a bottleneck.” - Nathen Harvey

DORA’s research confirms this: organisations that adopt AI without addressing process bottlenecks often see throughput decline and instability rise. The takeaway? Don’t just adopt AI, also fix the system behind it before you scale it. AI rewards operational excellence but punishes inefficiency.

Value Stream Mapping: the power of the mapping, not the map

To identify and eliminate those inefficiencies, Value Stream Mapping (VSM) has proven to be one of the most effective tools. VSM helps teams visualize every step a change must go through before reaching the customer. But its greatest strength lies in the shared discovery it creates.

The important thing about value stream mapping is the mapping, not the map.

Nathen Harvey

Lead of the DORA Research Program at Google Cloud

The exercise of mapping brings cross-functional teams, engineering, operations, product/business, security, into the same room. It sparks conversations, uncovers friction, and often leads to unexpected “aha” moments. As Nathen explained, it’s the connections we make as we build the map that has the most value.

Platform Engineering: scaling quality and consistency

The conversation also touched on Platform Engineering, another major focus in today's modern IT organisations. For teams beginning AI adoption, this often adds extra complexity and challenges—such as adapting workflows for AI and deciding which tools to use. Platforms can help reduce this complexity by ensuring stability and standardization.

A quality internal platform, is one of the most impactful enablers of AI-assisted delivery. It consolidates reusable capabilities—like CI/CD pipelines, observability, security controls, and AI tooling—so that product teams can focus on business outcomes rather than infrastructure.

Platforms act as an enabler for hyperautomation and business value, ensuring that acceleration doesn’t compromise quality. They should be treated as long-term products rather than temporary projects—built with a clear vision, user empathy, and continuous improvement in mind.

Key Insights to use AI in your SDLC

AI amplifies what exists, it accelerates strengths and exposes weaknesses.

Fix the system before scaling it. Broken processes only break faster under AI.
Value Stream Mapping matters, the mapping, not the map, creates insight and alignment.
Platforms enable hyperautomation. Treat your internal platform as a product with a long-term vision.
Culture drives performance. Empathy and learning remain the true differentiators.